Computer Networking Research Projects
Big-Data Networking: Use Multi-Core Routers for Bulk Data Transfers
As data centers are gaining more importance in the Internet, the need for bulk data (a.k.a. big data) transfers within and between data centers is increasingly important, Most end-to-end, or e2e, sessions are now going through a datacenter. More essentially, these big data transfers are crucial to the maintenance and backup operations of data centers.
Obviously, if a single path is used for data transfers, the aggregate e2e transfer rate will be limited to the bottleneck(s) on that path. Parallel data transfers can be spread over the network in a non-overlapping manner to improve the aggregate throughput. The downside is that such parallel transfers require multi-path routing capability. Recent work showed that multi-path routing is a successful approach to address the big data transfers. The key focus of these techniques is to diversify and spread the paths available to the e2e transport while satisfying various constraints such as delay or loss. The paths generated by these multi-path routing methods were adopted and TCP sessions were successfully parallelized with effective solutions. However, these e2e transfers are yet to utilize multi-core CPUs available in most routers.
We consider a "parallel routing" framework that explicitly considers multi-core routers and employs shortest-path calculations only. The basic idea is to virtually slice the router topology into "substrate" topologies and assign them to a separate router core, which calculates a shortest path on the assigned substrate.
Rather than solving the multi-path routing problem all at once, our approach transforms it into two sub-problems:
- Slicing out substrates from the router topology so that the collection of the shortest paths on each substrate diverse and non-overlapping e2e paths
- Calculate shortest paths on each substrate
Since the latter problem is already being handled in legacy routers, our approach can easily be adapted to current routers if the former problem is solvable. In one point of view, our approach transforms the multi-path routing problem into a topology/substrate generation problem.
- Soran, F. M. Akdemir, and M. Yuksel, Parallel Routing on Multi-Core Routers for Big Data Transfers, Proceedings of ACM CoNext Student Workshop, Santa Barbara, CA, December 2013.
Optical Wireless: Light to Complement Radio Spectrum
Mobile ad-hoc communication is starting to find real-world applications beyond its military origins in areas such as vehicular communications and delay-tolerant networking. As the RF spectrum is getting saturated by recent advances in wireless communications, enabling optical spectrum in wireless communications is a necessary revolution for the ultra-high-speed mobile ad-hoc networks (MANETs) of the future. This project explores the potential for free-space-optics (FSO), a.k.a. optical wireless, communications in the context of very-high-speed mobile ad-hoc and opportunistic networking.
This project introduces basic building blocks for MANETs using FSO and prototype multi-hop, high-capacity FSO building blocks and protocols operating under high mobility. 3-d spherical structures covered with inexpensive FSO transceivers (e.g., VCSEL and photo-detector pair) solve issues relevant to mobility and line-of-sight (LOS) management via availability of several transceivers per node. Such structures facilitate electronic LOS tracking (i.e., "electronic steering") methods instead of traditional mechanical steering techniques. The project also investigates reliability protocols as management of logical data streams through multi-interface FSO structures pose a major challenge.
By abstracting FSO directionality and LOS characteristics, the project explores issues relating to routing and localization, and develops layer 3 protocols and FSO-MANET demonstration in a lab setting. Results of this research can revolutionize the MANET technologies by enabling optical spectrum. FSO has been used at high-altitude communications, and this project enables FSO communications at lower-altitudes and in ad-hoc settings. This research will provide a new application for solid-state lighting technology due to potential integration of illumination and communication functions.
We have recently published results of our proof-of-concept prototype on the concept of "electronic steering" on a multi-transceiver node. The aim of the prototype is to illustrate that it is possible to seamlessly switch (i.e. steer) an ongoing data flow from one FSO transceiver to another without giving a significant disturbance to the ongoing data flow.
Figure 1: Front and rear views of our FSO transceiver with two LEDs and one PD. The transceiver diameter is 25mm.
We first designed a transceiver (shown in the figure at the left) composed of two infrared LEDs and a photo-detector (PD). Each transceiver has a serial port interface through which it is possible to modulate the LEDs and read the signal received at the PD. We placed the PD at the rear of the transceiver board to reduce the amount of optical feedback from LEDs.
Figure 2: A 3-transceiver circular node structure.
We then combined multiple transceivers on a circular structure (shown in Figure 2) and connected them to a breadboard microcontroller. We programmed the microcontroller so that a line-of-sight (LOS) alignment protocol is applied to detect availability of alignments on the transceivers and if so use those alignments to transmit data. The LOS alignment protocol probes availability alignments by periodically sending search frames from transceivers and uses three-way handshakes to assure bi-directional alignments with neighbor nodes.
The microcontroller also interfaces with a laptop via serial port. The laptop sends and receives data (e.g., an image or voice) via its serial port without knowing that a multi-transceiver FSO structure is being used to send or receive the data.
We assembled three such combinations of laptop, 3-transceivers and microcontroller and tested the possibility of seamless switching of data transmissions from one FSO transceiver to another while the node structures move with respect to each other.
Figure 3: Three laptops communicating via the multi-transceiver FSO node structures.
The experimental setup is shown in Figure 3. The three laptops establish two simultaneous separate data transmission (e.g. from A to C and C to B) and during these transmissions we move the nodes with respect to each other and observe that the electronic steering hands off the ongoing transmission to a new transceiver that is aligned with the neighbors.
We plan to migrate our prototype to Ethernet ports of the laptops and attain higher transmission rates with better quality FSO transmitters and PDs. The end goal of our prototyping efforts to realize an FSO-MANET utilizing several of such multi-transceiver nodes performing simultaneous data transfers among each other.
Further details and results of our prototype are available in our papers below.
As shown in Figure 4, we envision a spherical multi-transceiver optical antenna that will include many more transceivers and apply advanced learning and algorithmic methods to guide LOS alignment across its transceivers. With dense packaging of hundreds of transceivers on these node structures it will be possible to establish several simultaneous ongoing data transmissions with each neighbor separately. Further, by using the directionality of the FSO signals it will also be possible to detect angle-of-arrival and use it for relative localization, a concept that we explored with simulation-based experiments in our papers.
Figure 4: "Soccer ball" or "Fly's eye" 3-D optical antenna.
- M. Bilgi and M. Yuksel, Capacity Scaling in Free-Space-Optical Mobile Ad Hoc Networks, Ad Hoc Networks, Elsevier Science, Volume 12, pages 150-164, January 2014.
- Sevincer, A. Bhattarai, M. Bilgi, M. Yuksel, and N. Pala, LIGHTNETs: Smart LIGHTing and Mobile Optical Wireless NETworks - A Survey, IEEE Communications Surveys and Tutorials, Volume 15, Number 4, pages 1620-1641, April 2013.
- Sevincer, M. Bilgi, and M. Yuksel, Automatic Realignment with Electronic Steering of Free-Space-Optical Transceivers in MANETs: A Proof-of-Concept Prototype, Ad Hoc Networks, Elsevier Science, Volume 11, Number 1, pages 585-595, January 2013.
Network Management Game: Train Network Admins via Games
Online management of an active, large-scale network poses many challenges, which have attracted significant research.
As critical applications, such as high-definition TV (IPTV) and financial markets, are converging onto the Internet infrastructure, effective response to large-scale network dynamics like failures and demand spikes is gaining more importance. Link or node failures are not rare events for a large-scale network of thousands of devices. A major portion of the time spent handling such network dynamics is determining how to respond, mostly performed manually in the current practice.
Seeking the optimal response is often impractical, but even settling on a good response is very hard as well. The emergence of various networking technologies like 3G wireless and mesh networking is further complicating these management tasks. In most cases, getting the large-scale network to work is the typical target. Experienced human administrators are typically the ones who can quickly find a close-to-optimum response. However, as the networks are getting larger and more diverse, managing and attaining effective responses for an online operational network necessitates meta-tools to swiftly learn and characterize the network. This project responds to this fundamental need by developing tools to achieve automated ways of managing a running network.
The project develops tools for automated management of a running network by framing heuristic optimization, empirical learning, experimental design, and network management with a game interface. The project will develop an online management and experimentation system for large-scale networks in a game-like environment for trainee administrators to play with and explore what-if scenarios, without having to risk the network operation.
The project will also develop algorithms for empirical characterization of network dynamics, and tools for quick and close-to-optimal configuration of numerous network parameters in response to failures or customer traffic trends. Such a framework will automate the process of configuring a large-scale network, and thus reduce the dependency of ISPs to human network operators.
The project integrates behavioral-scientific concepts into the practice of operational network management. The automated management using online optimization may establish a foundation for managing multi-owner systems, e.g., power grid, transportation, and water infrastructure networks. The project's heuristic optimization and experiment design methods as well as the game-based approach to operator training are applicable to training in safety and mission critical industries where mistakes of ill-trained administrators are intolerable, e.g., airline pilot and nuclear reactor administrator training.
- E. Arslan, M. Yuksel, and M. H. Gunes, Network Management Game, ACM SIGCOMM Computer Communication Review, Volume 43, Number 1, pages 46-50, January 2013.
- M. Yuksel, K. K. Ramakrishnan, S. Kalyanaraman, J. D. Houle, and R. Sadhvani, Required Extra Capacity: A Comparative Estimation of Overprovisioning Needed for a Classless IP Backbone, Computer Networks, Elsevier Science, Volume 56, Number 17, pages 3723-3743, November 2012.
- B. Gonen and M. Yuksel, Network Configuration and Management via Two-Phase Online Optimization, Proceedings of IEEE Global Communications Conference (GLOBECOM), pages 1-6, Houston, TX, December 2011.
Network measurements help in modeling, understanding, and improving networks. Understanding the topological and the functional characteristics of the Internet is an important research issue. This understanding is not simply an intellectual curiosity but also a necessity in order to better design, implement, protect and operate the underlying network technologies, protocols, and services.
Internet measurements arise due to commercial, social, and technical issues and provide insight into network topology, routing, protocols, and applications. In this project, we focus on the Internet topology and the network traffic.
We are currently exploring efficient and comprehensive link-level Internet topology sampling. We try to identify graph sampling issues in various networks and provide means to reduce errors in the topology sampling practices. In addition, we look into mechanisms to detect presence of anomalies (e.g., link or router failure) or crisis situations (e.g., natural disasters or physical attack on the network infrastructure).
In addition to studying Internet topology, we analyze network traffic to understand the dynamic phenomena. By analyzing traffic patterns, we will be able to enhance underlying topologies, provision network resources, and design better protocols to meet the needs of emerging applications.
- Bingdong Li, Esra Erdin, Mehmet Hadi Gunes, George Bebis, and Todd Shipley, "An Overview of Anonymity Technology Usage", Computer Communications, 36(12): 1269-1283, 2013.
- Bingdong Li, Jeff Springer, George Bebis, and Mehmet Hadi Gunes, "A Survey of Network Flow Applications", Journal of Network and Computer Applications, 36(2):567-581, 2013.
Internet Topology Mapping
Internet topology research helps in understanding of macroscopic structure of the physical Internet structure, developing topology-aware algorithms, and providing simulation and topology generation tools for other research studies.
In this project we focus on collecting and analyzing link-level Internet topologies. In order to obtain sample Internet topologies, one needs to collect a large number of path traces and combine them. However, there are several challenges in obtaining accurate sample Internet maps from raw path traces. In particular, we have focused on resolving anonymous routers, alias IP addresses, and subnets.
We have built an Internet Topology Mapping System, named Cheleby, to construct accurate Internet maps from collected path traces. This system combines our work on the Internet topology measurements and will periodically provide Internet maps in addition to raw traces.
It is important to accurately and efficiently process raw path traces in Internet topology mapping before using the map in any study. Hence, we devise several mechanisms to accurately construct sample Internet topologies. The system will help in observing topological characteristics of the Internet and provide Internet atlas, which we plan to annotate with the geographic location of the routers.
- Hakan Kardes, Talha Oz, and Mehmet Hadi Gunes "Cheleby: A Subnet-level Internet Topology Mapping System", 4th International Conference on COMmunication Systems and NETworkS (COMSNETS), Banglore, India, Jan 3-7, 2012.
Social Network Analysis
Academic ties of countries.
As social networks have grown vastly in size and heterogeneity, complex network studies have become necessary and gaining popularity in recent years. Many researchers are formulating theory for the growth and the structure of the networks from different fields including biology, chemistry, geography, mathematics and physics. Complex network analysis helps to capture small scale and large-scale features of these networks that are not very obvious. Such analysis also may uncover the underlying dynamics of network growth.
In this direction, our research sets the following goals:
- Design friend recommendation systems by using complex network theory, cognitive theory and a Pareto-optimal genetic algorithm
- Model a software project as a network to identify the key authors and subject matter experts
- Analyze the collaboration of researchers when they obtain federal funding;
- Analyze academic ties of mathematicians using the Mathematics Genealogy Project data
- Examine the Sourceforge.net community of developers through the use of network analysis
- Seek to map changes in the news ecology of the San Francisco Bay area using network analysis
- Hakan Kardes, Abdullah Sevincer, Mehmet Hadi Gunes and Murat Yuksel "Six Degrees of Separation among US Researchers", IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2012), Istanbul, Turkey, 26-29 Aug, 2012.
- Adrienne E. Breland, Mehmet Hadi Gunes, Karen A. Schlauch, and Frederick C. Harris "Mixing Patterns in a Global Influenza A Virus Network Using Whole Genome Comparisons", IEEE Computational Intelligence in Bioinformatics and Computational Biology (CIBCB 2010), Montreal, Canada, May 2-5, 2010.
Cloud-Based Online Social Networking
As the Internet has become a cornerstone of our daily communications, people have been sharing their daily activities and opinions over the online social networks. These systems raise privacy concerns, however, as the users are not in charge of their content. Hence, we propose a cloud based online social network that will integrate cloud systems into peer-to-peer paradigms to provide higher availability and lower delay in sharing content between friends.
Our system will be app-based where a background process will exchange posts with online friends and keep a backup of the data on the cloud. The data will be encrypted before transmission to cloud environment that will help efficiently spread the user's content to his/her friends especially when the user is offline.
- Jeffrey Naruchitparames, Mehmet Hadi Gunes and Sushil Louis, "Friend Recommendations in Social Networks using Genetic Algorithms and Network Topology", IEEE Congress on Evolutionary Computation, New Orleans, LA, June 5-8, 2011.
Advanced techniques are needed to understand the underlying topologies of large complex networks. In this project, we introduce a novel way to visualize large-scale network topologies. We propose an Inner Sphere visualization method that projects the network topology on the inside of a sphere. User navigation around the network is accomplished through moving the sphere around the user's point of view.
Since previous research has shown that the spatial cognition ability in humans greatly affects the usefulness of a user interface, two empirical experiments were performed to test the usefulness of viewing topologies on a sphere compared to a flat surface. Our study indicated that network navigation on a sphere is faster but can also be confusing. Thus, we added more guidance tips to create a more intuitive user interface and to improve navigability.
Our Inner Sphere visualization method is implemented as a tool for interactive network visualization called GerbilSphere.
- David Shelley and Mehmet Hadi Gunes, "GerbilSphere: Inner Sphere Network Visualization", ComNet Complex Dynamic Networks: Tools and Methods, 56(3):1016-1028, Feb 2012.
With the Internet becoming central in daily communication, network security issues become very critical. Various communication paradigms have been deployed, each requiring specific security protection. In particular, as the Internet becomes a centerpiece of our daily lives, pirates are developing attack mechanisms using communication resources instead of traditional approaches.
In this project, we analyze the Internet traffic to gain more insight on attacks such as spam and worms. This understanding will guide the design of better communication mechanisms and protocols to prevent such attacks. In addition, we try to identify security flaws of emerging network technologies and develop attack prevention mechanisms to thwart malicious behavior.
- Jeffrey Naruchitparames and Mehmet Hadi Gunes, "Enhancing Data Privacy and Integrity in the Cloud", International Workshop on Security and Performance in Cloud Computing, Istanbul, Turkey, July 4-8, 2011.
Private Communication and Execution
One central concern in today's computing world is the privacy and integrity of sensitive data processed at remote systems. In this project, we propose a blind processing service using trusted computing mechanisms to provide improved privacy and integrity to its users.
Utilizing blind communication and execution services, a user can exchange sensitive information with a remote system via isolated processes whose execution environment and data is shielded from the rest of the system after ensuring the system has correct hardware, trusted computing base, correct credentials, and trustworthy state.
Our goal is to allow information exchange between system components with protection mechanisms against everyone including system administrators.
- Mehmet Hadi Gunes and Cansin Y. Evrenosoglu, "Blind Processing: Securing Data Against System Administrators", IFIP/IEEE International Workshop on Management of Smart Grids, Osaka, Japan, April 23, 2010.
As criminals utilize advanced technologies, law enforcement will be at a disadvantage in staying well informed of the complex technical issues that accompany investigating and prosecuting high technology crimes. In this project, we investigated mechanisms track cyber criminals when they use anonymizer technologies. We determined potential exploitation of criminal systems and developed a tool suite that can automate the exploration process.
In this direction, we investigated:
- Local/global monitoring systems for major anonymizer networks
- Multiplication attack to correlate entry and exit nodes in anonymizer networks
- Data mining tools to correlate user behaviors so that flows can be classified
- Circuit Clogging techniques to insert busy cycles that can identify flows
- Watermarking techniques to identify users
In particular, we investigated how browsers can be used to correlate criminal activities of users, how application-level protocols such as bit-torrent can leak identifying information, and how document scripts can report identifying information.
Cognitive Radio Networking
The FCC's current initial proposal of opening sub-900 MHz TV bands to unlicensed services because of high under-utilization of these bands has created a new horizon of research activities. For unlicensed devices to gain access to these bands, it is mandated that these devices detect licensed users and avoid interference. The newly proposed Dynamic Spectrum Access (DSA) networks (e.g., IEEE 802.22) based on cognitive radios are seen as the solution to this current problem.
In this research, we investigate how to implement embedded cognitive radio systems that can learn from the environment and configure the transmission/reception parameters automatically to operate at any unused frequency band dynamically. The focus is on enhancing DSA networks and architecture, air-interfaces (PHY and MAC) and the feasibility of solving problems like self-coexistence and hidden incumbents.
Studying the evolution of DSA networks and their survivability under unintentional/intentional disruption from hostile/malicious users based on inter-disciplinary foraging theory and anthropological human-societies is one of the prime focuses of this research.
- K. Hong, S. Sengupta and R. Chandramouli, "SpiderRadio: A Cognitive Radio Implementation using IEEE 802.11 Components", IEEE Transactions on Mobile Computing, 2013.
- Shamik Sengupta and K.P. Subbalakshmi, "Open Research Issues in Multi-hop Cognitive Radio Networks", IEEE Communications Magazine, 2013.
- Shamik Sengupta, Swastik Brahma, Mainak Chatterjee and Sai Shankar N, "Self-coexistence Among Interference-aware IEEE 802.22 Networks with Enhanced Air-interface", Pervasive and Mobile Computing, 2013.
Dynamic Spectrum Access and security
Cognitive radio technologies are emerging as a platform to solve the problem of "artificial spectrum scarcity" for wireless applications; however, as with many new technologies, initial research has not focused on security aspects of cognitive radio networking from different classes of attacks from malicious nodes/users. The attack may happen in any of the following forms:
- Aggregation/bonding/fragmentation jamming (malicious intention of distorting normal secondary communications using the DSA features of aggregation or bonding or fragmentations of spectrum bands)
- Emulating characteristics of primary users of the spectrum band (malicious intention of evacuating the secondary users from the spectrum band and capturing the band) an
- Emulating characteristics of secondary users and launching various sybil attacks
Thus sensing/detection/forensics of attacks from malicious users is an important task to build efficient and secure DSA system. Security against these classes of attacks cannot be achieved through cryptographic means alone. In this research, we explore ideas from behavioral models, forensics and stochastic learning to optimize decisions under uncertainty and implement new cross layer optimization security protocols for dynamic spectrum access networks.
- Yi Tan, Shamik Sengupta and K.P. Subbalakshmi, "The Primary User Emulation Attack in Dynamic Spectrum Access Networks: A Game Theoretic Approach", IET Communications, 2011.
- Yi Tan, Shamik Sengupta, and K.P. Subbalakshmi, "Analysis of Coordinated Denial-of-Service Attacks in IEEE 802.22 Networks", IEEE JSAC Special Issue on Cognitive radio Networking and Communications, vol. 29, issue 4, pp. 890-902, 2011.
- Shamik Sengupta, Santhanakrishnan Anand, Kai Hong and Rajarathnam Chandramouli, "On Adversarial Games in Dynamic Spectrum Access Networking based Covert Timing Channels", ACM Mobile Computing and Communications Review (MC2R), Special Issue on Cognitive Radio Technologies and Systems, 2009.
To harness the wide variability of coverage, bandwidth, and reliability offered by different technologies, wireless service providers are already deploying heterogeneous access networks in an over-laid fashion operating in multiple spectrum bands. These heterogeneous networks would be capable of providing different sets of services governed by their corresponding quality-of-service (QoS) capabilities.
The current practice of offering different contract plans (based on minute usage) for voice services will no longer be valid for data services. The same notion of resource sharing in voice networks cannot be used because packet data systems are usually aimed at maximizing the throughput. This new way of looking at different charging for different QoS enabled services brings forth the requirement for a proper pricing model in dynamic spectrum access networks with heterogeneous access networks and differentiated services.
How will the WSPs determine the optimal price per unit bandwidth and advertise the same to maximize their market share? Note that, with multiple heterogeneous networks, the assignment of bandwidth over each network depends on the requirements of the admitted traffic. What will be strategy for adaptive bandwidth assignment? What will be the end-users' strategy in terms of WSP and network selection for a desired service?
- Y. Tan, S. Sengupta and K.P. Subbalakshmi, "Human Society Inspired Dynamic Spectrum Access Networks: The Effect of Parochialism", IEEE GLOBECOM, pp. 1-5, 2011.
- Yi Tan, Shamik Sengupta, and K.P. Subbalakshmi, "Competitive Spectrum Trading in Dynamic Spectrum Access Markets: A Price War", IEEE Globecom, pp. 1-5, 2010.
- Shamik Sengupta, Santhanakrishnan Anand, Mainak Chatterjee and Rajarathnam Chandramouli, "Dynamic Pricing for Service Provisioning and Network Selection in Heterogeneous Networks", Elsevier Physical Communication (PHYCOM) Journal, Special issue on Cognitive Radio: Algorithms & System Design, vol. 2, pp. 138-150, 2009.